R SIG Debian - Sep 2012 - Refresh times very slow for simple plots.

Hello everybody,

On my system (R 2.15.1-4 on Debian Stable plus some updates), resizing graphic
windows (by stretching their borders with the mouse) takes one order of
magnitude more time than redrawing them (by running the command again).

For instance, the command ??plot(data.frame(rnorm(10), rnorm(10), rnorm(10)))??
is executed almost instantely, but if I resize the graphics window, it takes 3
to 4 seconds to refresh. It looks related to drawing the lines making the
frames, the axis and the ticks, for instance ??heatmap(cor(data.frame(rnorm(10),
rnorm(10), rnorm(10))))?? is also strongly affected.  Even if each dendrogram is
only made of five segments, I can see them being plotted one after the other.

This becomes very handicaping when working on real-life datasets, as a graph
that takes a few seconds to draw will block my session for a few minutes if I
resize it by mistake.

I was wondering if somebody else was affected by this strange bug ?

As per the R FAQ here is the output of sessionInfo()

R version 2.15.1 (2012-06-22)
Platform: x86_64-pc-linux-gnu (64-bit)

locale:
 [1] LC_CTYPE=en_GB.utf8       LC_NUMERIC=C             
 [3] LC_TIME=en_GB.utf8        LC_COLLATE=en_GB.utf8    
 [5] LC_MONETARY=en_GB.utf8    LC_MESSAGES=en_GB.utf8   
 [7] LC_PAPER=C                LC_NAME=C                
 [9] LC_ADDRESS=C              LC_TELEPHONE=C           
[11] LC_MEASUREMENT=en_GB.utf8 LC_IDENTIFICATION=C      

attached base packages:
[1] splines   stats     graphics  grDevices utils     datasets  methods  
[8] base     

other attached packages:
[1] gdata_2.8.0   edgeR_2.6.1   limma_3.12.1  vegan_2.0-3   permute_0.7-0

loaded via a namespace (and not attached):
[1] grid_2.15.1    gtools_2.6.2   lattice_0.18-8

Have a nice day,

-- 
Charles Plessy
Tsurumi, Kanagawa, Japan

Yes, I suffer from that too, especially as I mostly have the x11() windows
pop up on a machine different from the one doing the calculation (the latter
being a headless server).

If you look at help(x11), you see that there are (at least) three device
types. At one point I made 'nbcairo' (non-buffered cairo) the default
and it
seemed to help.

The real culprit, though, is the R device model. And a real alternative is
offered eg by Simon's 'xplots extremes' using OpenGL acceleration,
see
http://rforge.net/Acinonyx/

He gave absolutely stunning live presentations using this. May be a good
alternative for 'large' data.

Dirk

-- 
Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com